Unhobbled Donors Needed: AI Safety Funding Must Front-Load Before Capital Wave

A large wave of philanthropic capital is headed toward AI safety, but it will arrive slowly and unevenly. In a new LessWrong post, Saheb Gulati warns that the window for high-leverage intervention is closing. Political timelines—especially the 2026 midterms and the Trump administration’s emerging AI stance—mean that money spent in 2026 can shape policies that later funding cannot influence. Talent pipelines also need time: early grants can recruit and onboard smart young people before they commit to capabilities work. Institutional credibility takes years to build, and first-mover funders can define the frameworks and paradigms (e.g., race dynamics, nonpartisanship) that guide later debate. Gulati calls these fast, risk-tolerant donors “unhobbled” and argues they are the missing category in the AI safety ecosystem.

Unlike traditional philanthropists who wait for proven interventions, unhobbled donors deploy capital before the wave, bet on early-stage projects, and publicly campaign for neglected causes. They accept that the cost of giving too early is small compared to the cost of giving too late, especially if timelines are short. By seeding projects that megafunders will ignore and setting agendas that later money follows, these donors can achieve outsized impact. Gulati urges individual donors and small grantmakers to radically rethink priorities, timelines, and risk tolerance—because the leverage available now will not be available in 2028.